Siple Coast ice streams reorganization following the reactivation of Kamb Ice Stream tributaries, West Antarctica

Ice streams in Antarctica account for most of the ice volume discharged to the ocean, and their flow variability greatly influences the mass balance of the ice sheet. Today, the Siple Coast region is the only one in West Antarctica to experience a positive mass balance of 36 Gt/yr, as a consequence of the stagnation of Kamb Ice Stream 170 years ago as well as the ongoing slowdown of Whillans Ice Stream. However, this positive trend could be temporary, as reconstructions of past flow show that the Siple Coast ice streams have experienced significant flow variability over the past millennia, with episodes of stagnation typically followed by reactivation on a centennial timescale. The impact that this internally induced variability may have on the future mass balance remains unknown. Here, we use the three-dimensional ice sheet model CISM to study the flow variability of the Ross ice streams over a 250 year period, during which the basal properties controlling ice flow evolve according to the distribution of meltwater in a dynamic, regional hydrological system, coupled to a subglacial till layer. To obtain realistic initial flow conditions, we assimilate the 1997 velocity data available for this region. During the forward simulation, all ice streams undergo dynamic changes, traced to persistently weak bed conditions beneath the tributaries of the currently dormant Kamb Ice Stream. This source of regional instability propagates into the catchments of Whillans as well as Bindschadler ice streams within decades, and re-organizes the flow of all the Ross ice streams within two centuries. Ice piracy from Kamb Ice Stream prolongs the fast flow on Whillans Ice Stream, maintaining its trunk active during the first modeled century. At the same time, ice piracy into the catchment of Bindschadler ice stream reactivates the relict Siple Ice Stream, which significantly reduces the mass balance for this region. The trunk of Kamb Ice Stream reactivates last (during the second half of the experiment), because high rates of basal freezing were induced after its stagnation. These model outputs are similar to the reconstruction of past flow. We discuss the mechanisms of flow migrations, and the implication for the future mass balance of the region.